**This text was part of an extinct chapter of Visual Complexity: Mapping Patterns of Information, which never saw the light of day. Instead of being forgotten in a dusty folder, I decided to make it available to the general public and invite any constructive criticism by our growing community. Hope you will find it useful.**
Data and information visualization are fundamentally about showing quantitative and qualitative information so that a viewer can see patterns, trends, or anomalies, constancy or variation, in ways that other forms – text and tables – do not allow.
The concept of visualization is certainly not new. Humans have been involved in the visual representation of information for more than 30,000 years. During this time, there has been a variety of portrayed subjects, many of them pertaining to natural phenomena, but the common underlying purpose of communicating a message has always been present. Whether we talk about cave paintings, cuneiforms, maps, or charts, we are always alluding to information in a quality of a message from a sender to one or more receivers. “The progress of civilization can be read in the invention of visual artifacts, from writing to mathematics, to maps, to printing, to diagrams, to visual computing.”, say Card, Mackinlay and Shneiderman. Historian Alfred W. Crosby attests to the importance of visual aids throughout the ages, by claiming that visualization and measurement were the two factors most responsible for the rapid development of all of modern science.
Even though visual artifacts have always been a central element in the history of humankind, over the last 25 years the term “visualization” has become immensely popular, being fragmented in a profusion of subfields, carrying a diversity of specialized labels such as Information Visualization, Data Visualization, Scientific Visualization, Software Visualization, Geographic Visualization, Knowledge Visualization, Flow Visualization, and even Music Visualization. Many of these areas emerged in the midst of existing parallel areas like Information Design, Information Graphics, and Visual Communication. The distinction between them is occasionally thin, and in some cases almost inexistent. This rich plethora of labels is certainly indicative of the outburst of a new practice, but one that is still struggling to define itself. While some consider this to be the birth of a new medium, or even a new science, the consensus on a definite descriptive label is not so obvious.
According to Michael Friendly, the renowned professor of Psychology at York University in Canada, information visualization is the broadest term that could be taken to include all the developments in visualization, since “almost anything, if sufficiently organized, is information of a sort: tables, graphs, maps and even text, whether static or dynamic, provide some means to see what lies within, determine the answer to a question, find relations, and perhaps apprehend things which could not be seen so readily in other forms.” But even able to accommodate the broadest of scopes, information visualization has also been the definite title of a multidisciplinary field emerging out of the computer science community in the late 1980s.
Originally coined by Jock Mackinlay and his User Interface Research Group at Xerox PARC in 1986, information visualization relates to the “use of computer-supported, interactive, visual representations of abstract data to amplify cognition”. It’s in essence a computer-driven transformation of abstract data (distinct from physical data – the earth, molecules, cells, human body, etc) into an interactive visual depiction aiming at insight – which in turn translates into “discovery, decision-making, and explanation”. Congregating a vast body of research from computer science, human-computer interaction, communication design, cognitive psychology, semiotics, statistical graphics, cartography, and art, modern information visualization surfaced from advances in computer graphics and was further consolidated in 1987, when the NSF Panel on Graphics, Image Processing, and Workstations published its landmark report Visualization in Scientific Computing. Since then, information visualization has grown considerably as an independent discipline, fostered by many conferences and workshops dedicated to the topic, particularly the prominent IEEE Computer Society symposium on Information Visualization, known as the InfoVis conference, first held in 1995.
With roughly two decades, information visualization has already been the target of some criticism and dismissal. Most of it comes from an inadequacy of the field to swiftly adapt to recent changes, caused by a large adoption from eager art and design communities and an escalating curiosity from media, advertising, and publishing. As a close-knit group, naturally inclined towards the computer science community, as a result of its own heritage, information visualization must take a stance to either adjust to these changes and fully accept its growing popularity, or instead, remain a niche inward-looking academic practice. Some signs of an embrace between traditional circles and the new wave of enthusiasts are already starting to surface, and this initial hesitation might simply go down in history as the normal shyness of a first date. Nonetheless, it is not surprising that under the present uncertainty, some voices have come forward suggesting new terms and definitions. Ben Fry in his PhD thesis defended a new label called “Computational Information Design”, able to properly integrate information visualization, data mining and graphic design, while Robert Kosara is a promoter of “Visual Analytics”, with a stronger emphasis on analytical reasoning. While many of the arguments for new labels reinforcing specific scientific or design concerns are certainly valid, there’s a major concern of an excessive breakup of a field that’s still defining itself.
Instead of trying to devise new titles for alternative branches highlighting a particular area of focus, the effort should be in creating a bridge between the existing body of research and the abundance of novel demands, in an attempt to revise and renovate the field, steering information visualization into a mature, integrated, and in demand hotspot. If willing to adapt, the field is broad enough to fully encompass most requirements, from a stronger prominence of design to a reinforced attention to analytics. This doesn’t mean the discipline can incorporate any attempt at visualizing data. But in essence, all interactive visual representations, able to make the depicted subject more intelligible and transparent, or find a new explicit insight within it, can and should be embraced by information visualization.
Information visualization is well known for its multidisciplinary nature, assembling people from a vast assortment of backgrounds, but notwithstanding the contribution of innumerous disciplines, we can still highlight three main spheres of activity that best characterize its key attributes and capabilities. Readers familiarized with research publications in the field will find this conception slightly different from previous frameworks developed by Stuart Card, Jock Mackinlay, Ben Shneiderman, and Ed Chi. The deliberate intent of this reframing is to emphasize the leading role of design, in both visual and interactive choices, and the fundamental function of statistics and data mining. This is ultimately an integrating, yet diverse framework, keeping alive the heterogeneous nature of the discipline. Here we describe the three central layers of information visualization: Data Transformation, Visual Mapping and Interactive Framing. Even though there’s a natural progression between the three stages that doesn’t mean they sustain in a fixed order. There’s a lot of refinement taking place in a continuous iterative process that forces each step to be occasionally revisited.
This is the very first stage in the development of any information visualization project. Without data no visualization would even be possible, hence everything starts by attaining access to a particular dataset relevant to the project’s pursuit. After getting hold of the data, what follows is a long process of data analysis, which includes inspecting, cleaning, filtering, and parsing the data, while organizing the relevant parts and removing the irrelevant. The subsequent process of data mining is crucial in order to have a better understanding of the natural affordances of the dataset. It encompasses a series of queries and algorithms in order to extract particular patterns in the data for some quick modeling and visualization tests, which will be of great importance in the build up of the second stage. Data transformation is the essential foundation of a successful execution, and covers areas like programming, statistics, data analysis, data mining, analytics, and machine learning.
Visual mapping is a critical step in information visualization, where data finally comes to life through a deliberate visual form. It takes into consideration key factors like top-to-bottom hierarchy, color, legibility, typeface, contrast, spacing, position, size, shape, orientation, layout, and depth. This central task contemplates not only individual views or modules, but also the composition of the entire contiguous environment. The choice of a particular method (or methods) is tied with the specific goal of the piece – its intrinsic purpose – and might be defined a priori or during project development, as the natural affordances of the data come into place. It’s also highly dependent on end users, their immediate context and expressed needs – when, where, and how the final execution will be used. Visual mapping is tied with various areas of visual design, including graphic design, information design, interface design, visual perception, cognitive psychology, aesthetics, and typography. Furthermore, it’s essentially made of two components: graphical objects and textual objects.
Information visualization is ultimately a discovery tool, and interactivity provides the final coalescing layer for exploration. “Visual representations and interaction techniques take advantage of the human eye’s broad bandwidth pathway into the mind to allow users to see, explore, and understand large amounts of information at once”, elucidate James Thomas and Kristin Cook, and they further explain, “Visual representations alone cannot satisfy analytical needs. Interaction techniques are required to support the dialogue between the analyst and the data. While basic interactions such as search techniques are common in software today, more sophisticated interactions are also needed to support the analytical reasoning process.”
Some don’t see the clear-cut need for interaction in information visualization, so it’s important to clarify this assertion. In a broader definition of visualization, it’s broadly consensual that information can be successfully conveyed in either static or interactive executions. However, we have to question what really sets information visualization apart from other parallel fields such as information design or information graphics. It’s in fact its computer-supported interactive nature that truly makes it distinct, and this unique offering becomes imperative as the degree of complexity of the portrayed system increases. The representation of complex networks is just an instance where interactivity is vital. Coupled with a relevant time-variant dataset, interactivity can also be a critical driver in a shift from short-term casual engagement to long-term active engagement, substantiating information visualization as a significant tool for exploration.
But interactive framing is not limited to the constraints of a computer screen. It covers any responsive visualization where a two-way communication between user and layout is established, from reactive surfaces to highly immersive visualization environments. This ultimate unifying layer is critical for explorative analysis, enabling users to inquire, filter, manipulate, reshape, and examine the visual outcome in order to identify properties, relationships, regularities, or patterns. Finally, it’s important to elucidate that even though interactivity is a central component of information visualization, the field doesn’t aim at replacing static depictions of information, since they can successfully complement each other. It simply provides an alternative, yet extremely powerful medium.
Even though there is a widespread consensus on its qualifications, information visualization, as a recent emergent field, still lacks a structural foundation able to uphold and expand its projection well into the future. We cannot consciously claim to be a new medium or a new science, when innumerous questions are still unresolved. It is critical for such an introspection effort to happen without delay, since there’s too much work to be done, and once we all agree on what we do as a community, it will be easier for external parties to recognize the goals and boundaries of our discipline. It’s obvious that we are still pulling together the different parts that make this practice and trying to understand when best to use them, but in order for information visualization to take the next step, and grow into a cohesive field of study, it requires the consolidation of three critical components:
Assemble a clear underlying theory able to combine many of the learnings, knowledge and insights from the variety of disciplines that make information visualization. If recent years have been marked by a significant profusion of new projects, this sturdy practice needs to be sustained by a reliable system of ideas and ideological principles. The purpose is to ultimately provide a broad consensual framework able to evaluate past, present and future endeavors. The current unguided exploration is by no means detrimental, since it’s the perfect setup for innovation to sprout, however, if the discipline wishes to mature as a reliable knowledge domain, it needs a supporting body of theory capable of accommodating all recent advances. Cognitive psychology might be one of the most reliable instruments in the edification of such a system, able to easily translate cognitive behaviors into objective design principles. A theory of information visualization will have to embrace diversity, and consequently several theories might need to coexist in opposition to one universal all-encompassing framework.
Define the spectrum of representational methods and techniques of information visualization. The central aim should be to consolidate and further exemplify, by recognizing the different data types and structures that underlie a common typology of patterns. Chaomei Chen, an important figure in the field, asserts on this current call to arms: “a taxonomy of information visualization is needed so that designers can select appropriate techniques to meet given requirements”. This is not meant to be a fixed and definite taxonomy, but an evolving, ever-growing, ever-expanding endeavor. This effort doesn’t contemplate a mere collection of techniques either; it should foremost supply a set of foundational principles able to guide present and future practitioners. Some initial steps in the description of common information visualization patterns have started to arise, but we still have a long way to go.
Provide easy evaluation methodologies for existing tools and approaches. Information visualization requires a common rule system that can accordingly distinguish the good from the bad, the appropriate from the inappropriate, the usable from the unusable, the effective from the ineffective. Case studies and success stories are a great first step in this direction. If information visualization is a vehicle for evidence and clarity, it should embrace the same ideology in the definition of its own practice, by creating a systematic body of analysis able to properly evaluate the success of any project. Quantitative and qualitative evaluation methods should be welcomed, including observational studies, participatory assessment, usability testing, contextual interviews, and user feedback. This effort should, most importantly, go hands in hands with the development of an adequate language of criticism.
Friendly, Michael. “Milestones in the history of thematic cartography, statistical graphics, and data visualization”. August 24, 2009. http://www.math.yorku.ca/SCS/Gallery/milestone/milestone.pdf (accessed October 23, 2011).
 Card, Stuart K., Jock Mackinlay, and Ben Shneiderman. Readings in Information visualization: Using Vision to Think, 5.
 Crosby, Alfred. The Measure of Reality: Quantification and Western Society, 1250-1600.
 Friendly, Michael. “Milestones in the history of thematic cartography, statistical graphics, and data visualization”, 2.
 Card, Stuart K., Jock Mackinlay, and Ben Shneiderman. Readings in Information visualization: Using Vision to Think, 6
 Thomas, James J. and Kristin A. Cook. Illuminating the Path: The R&D Agenda for Visual Analytics, 30.
 Chen, Chaomei. Information visualization: Beyond the Horizon, 1.